It is great to have more solutions but I never understood what was the point to distinguish such kind of projects from the commercial alternatives? Access to the source code is doubtfully an advantage. They charge for their product and in order to use it you have to pay. Of course, there will be people who will or willing to work for free to create an open source end user solution but IMO it is insignificant part of the market. Moreover they have similar prices to the established vendors products. With the same price tag will you purchase ALPR from Genetec, ipConfigure or them just because you can see the source code?
One thing about this model that I personally like is that you could (as an integrator or manufacturer) set it up and test it on your own.
As an integrator, you might not have many customers that ask for LPR, but it would be good (IMO) to be familiar with a solution. For most of the alternatives you would have to pay out of your pocket, or ask for a discounted demo unit.
I like the idea of setting something like this up and just letting it run on a demo cam at my home or office. Watch how it works, see what environment I need to get the best results. Then when/if a customer asks me for an LPR solution I at least have one thing I am familiar with.
If you don't do many LPR jobs the price may be less of an issue. If you're doing a lot of LPR, then it might make sense to go with a "bigger" product, one that is likely to have more developers, more support staff, etc.
One of our customers in Toronto chose OpenAlpr over another US based LPR solution due to their free trial software and superior support. The integrator setup a system at their office and tested it under variable lighting conditions (day, night, rain etc) and with a few different camera models and optics choices. The solution they deployed required a custom integration with ICT's ProtegeGX access control system and IVT Vision VMS. Vehicle plates entering a parking facility would be logged along with the plate number of the vehicle, the user credential along with an image of the driver and an image of the actual vehicle.
So I have been playing around with it this afternoon (just the API/CLI) free version) its not bad. you do need to make sure everything is configured properly and have a really good angle on your camera for the best results.
I will say if you stream straight from the camera live stream I have only been able to get it to work off of MJPEG shots as it doesn't seem to support raw .h264/rtsp( which makes sense) but is kinda deceptive from the documentation.
I had it running on some old and busted desktop hardware and it seemed to run rather well considering. I am awfully interested in the GPU acceleration option as if it works you could cut down on hardware cost tremendously. sadly I dont have anything available to test it at the moment
At the consumer level, it's likely a small but not insignificant minority that respond well to this marketing approach.
At the developer level, there is an immense investment in learning and becoming effective within a system. Since a rising tide can lift all boats, this is less relevant within a strong and supportive vendor ecosystem. Eventually, if partners disappoint, a business' future can critically depend upon elements beyond their control. This is one reason for the attractiveness of an open infrastructure.
IPVM has a great breadth of scope, which brings us a relevant use case that resurfaced within a few hours of the question:
"The program included a contract for Microsoft to build a PSIM and datamining tool, which he says lacked the usability he would have preferred.
One of the main problems with the Microsoft system was the lack of customization and control. If the city wanted to add more data sources, it meant going through the vendor. There was no way to make changes and customize the software without calling the vendor back."
The open source /free version only supports MJPEG while the commercial version supports both MJPEG and H.264. The commercial OpenALPR Agent has a number of enhancements. With these enhancements, the commercial agent is significantly faster when analyzing video streams.
Additionally, there are a number of features that are available exclusively to commercial customers.
then you have failed to keep your repos straight because the copy I got from github a while back lets me do everything but use the GUI and process live RTSP streams from the command line you might want to "take care" of that for future endeavors.
SimpleLPR worked better for me. The price was less, my application is working with up to 4 cameras on a server. It processes MJPEG streams. I can detect the license plate and send it to Genetec SC where I get it with a TCP server macro, and add as bookmark, or even I can parse a blacklist/whitelist text file for match and control relays, etc. If you dont need expensive highend solution then its a good low cost LPR for smaller or even mid-sized systems.
Does SimpleLPR process MJPEG streams? SimpleLPR's manual lists support for still image formats: JPEG, TIFF, PNG and BMP. Also, SimpleLPR does not support USA plates and H.264 video. As a forensic tool it would be somewhat useful but you can save $450 by using OpenALPR's open source for non-commercial use and accomplish the same results.
" it would not be difficult to feed data into a POS interface for a VMS"
Can you clarify what "POS" means in this instance?
This is better than some of the other cloud solutions I have seen which requires the camera to have Internet access, which isn't always the case. Having the agent run on a computer, be it the VMS computer or a separate one that has access to the camera network and also the Internet, is much more universal and flexible.
Many VMSes have the ability to take in a text feed that is associated with the video. Some layer the text over the video, some display it in a secondary viewing window, and some index the text and make it searchable.
The most common text source is from a Point Of Sale (POS) device, so that data about the transaction that was rung in by the cashier can be seen alongside the video. This helps loss prevention people spot cashiers ringing up "No Sale" or sweethearting stuff for their friends.
For this example, I am talking about having a camera feed into an OpenALPR instance to detect plates, and then having OpenALPR output the data it extracts from the image into the POS interface of a VMS. This would potentially make the video searchable for license plates that have come through.
OpenALPR introduced new Cloud Stream service plans, including a free plan and lowered the price of its monthly service. We have also introduced an Axis camera Agent and ALPR as a Service (no VMS, no server, no annual contract required). See http://www.openalpr.com/cloud-stream.html for details.